A retailer with 2,000 stores needs forecasts per geographic area that account for holidays and weather. They want to test the model for 2–3 days and need it to adapt to supply chain and store constraints. Which combination of steps provides the required functionality with the least operational overhead? (Choose two.)

The retailer needs forecasts per geographic area that account for holidays and weather, which are directly supported by Amazon Forecast's built-in holidays featurization and weather index. This minimizes development effort and operational overhead compared to manually implementing these features. For testing, a canary rollout allows for short, controlled tests (2-3 days) on a small subset of stores before full deployment. SageMaker can host the model, and Step Functions can orchestrate the canary deployment logic, managing the gradual rollout and monitoring. A/B testing is typically for comparing two models over a longer period, not for short, adaptive tests. Developing with Prophet in Amazon Forecast is an option, but the question specifically asks for holiday and weather accounting, which are distinct features in Forecast.